Benchmarking Robustness of 3D Point Cloud Recognition Against Common Corruptions

2022/01/28 by J. F. Sun, Jiachen Sun, Qingzhao Zhang +11 · 49 citations
Computer Science · Earth and Planetary Sciences · Engineering · #3D Surveying and Cultural Heritage #Advanced Neural Network Applications #Artificial Intelligence (cs.AI) #Artificial intelligence #Benchmarking #Cloud computing #Computer Vision and Pattern Recognition (cs.CV) #Computer engineering #Computer science #Data mining #FOS: Computer and information sciences #Industrial Vision Systems and Defect Detection #Machine Learning (cs.LG) #Machine learning #Operating system #Point cloud #Robustness (evolution) #cs.AI #cs.CV #cs.LG

paper · pdf · doi:10.48550/arxiv.2201.12296

published in arXiv (Cornell University) (Cornell University) · Codebase and dataset are included in https://github.com/jiachens/ModelNet40-C

arxiv created 2022/01/28 · openalex publication_date 2022/01/28 · arxiv updated 2022/01/31 · openalex created_date 2022/05/05 · openalex updated_date 2026/08/05

Abstract

Deep neural networks on 3D point cloud data have been widely used in the real world, especially in safety-critical applications. However, their robustness against corruptions is less studied. In this paper, we present ModelNet40-C, the first comprehensive benchmark on 3D point cloud corruption robustness, consisting of 15 common and realistic corruptions. Our evaluation shows a significant gap between the performances on ModelNet40 and ModelNet40-C for state-of-the-art (SOTA) models. To reduce the gap, we propose a simple but effective method by combining PointCutMix-R and TENT after evaluating a wide range of augmentation and test-time adaptation strategies. We identify a number of critical insights for future studies on corruption robustness in point cloud recognition. For instance, we unveil that Transformer-based architectures with proper training recipes achieve the strongest robustness. We hope our in-depth analysis will motivate the development of robust training strategies or architecture designs in the 3D point cloud domain. Our codebase and dataset are included in https://github.com/jiachens/ModelNet40-C

Cited by

• Adapt-As-You-Walk Through the Clouds: Training-Free Online Test-Time Adaptation of 3D Vision-Language Foundation Models
• TACO-Net: Topological Signatures Triumph in 3D Object Classification
• Skeleton-based Robust Registration Framework for Corrupted 3D Point Clouds
• Desensitizing for Improving Corruption Robustness in Point Cloud Classification through Adversarial Training
• Joint graph entropy knowledge distillation for point cloud classification and robustness against corruptions
• Purge-Gate: Backpropagation-Free Test-Time Adaptation for Point Clouds Classification via Token Purging
• Deep Learning For Point Cloud Denoising: A Survey
• ModelNet40-E: An Uncertainty-Aware Benchmark for Point Cloud Classification
• 3D Test-time Adaptation via Graph Spectral Driven Point Shift
• OmniVec2 -- A Novel Transformer based Network for Large Scale Multimodal and Multitask Learning
• Exploiting Vision Language Model for Training-Free 3D Point Cloud OOD Detection via Graph Score Propagation
• Hard-Soft Pseudo Labels Guided Semi-Supervised Learning for Point Cloud Classification
• DriveSuprim: Towards Precise Trajectory Selection for End-to-End Planning
• SMART-PC: Skeletal Model Adaptation for Robust Test-Time Training in Point Clouds
• RECON: Robust symmetry discovery via Explicit Canonical Orientation Normalization
• Test Time Adaptation Methods for Point Cloud Registration in Laparoscopic Surgery

Related

• Benchmarking and Analyzing Point Cloud Classification under Corruptions
• Benchmarking and Analyzing Robust Point Cloud Recognition: Bag of Tricks for Defending Adversarial Examples
• CSI: Enhancing the Robustness of 3D Point Cloud Recognition against Corruption
• RobustPointSet: A Dataset for Benchmarking Robustness of Point Cloud Classifiers
• Desensitizing for Improving Corruption Robustness in Point Cloud Classification through Adversarial Training